l**********o
发帖数: 9952 | 1 我没怎么学过statitics，可能问题很傻，但现在没人可以问，大包子请教（有答就有包子，答的多答的好的包子大）：
我有一组8个数据，say (1.00,1.05,1.10,0.97,0.82,1.03,0.99,0.95)
我想做power analysis, to calculate the required sample size to detect a 10% change (is this termed as effect size?) with 95% confidence (is this termed as significance criterion?) with power of 0.8 (is this usually just fixed at 0.8?)
说得也许不清楚，换言之，我这8个是已测量的normal group，是一个group里的，用来作为normal mean and standard deviation 这些标准的.
We want to determine the number of patients and controls (assuming the same number of participants in disease group and the control group, i.e. sample size) that would be required for future experimental designs using the proposed method to detect effect size=10%, with confidence=95% and power=0.8。
请教有什么软件可以让我输入我的8个数据和以上这些的paramters算出这个required sample size的?
如果谁已经有这样的code（我最熟悉matlab)能给我用就太棒了！或者别人写好的free software也太好了。
还有，我上面这些对问题的描述有明显很外行的错误吗？
多谢！ | B******y
发帖数: 9065 | 2 你都已经有数据了，Sample Size＝8，你还需要做什么Power Analysis？ | l**********o
发帖数: 9952 | 3 我这8个是已测量的normal group，是一个group里的，用来作为normal mean and standard deviation 这些标准的.
We want to determine the number of patients and controls (assuming the same number of participants in disease group and the control group, i.e. sample size) that would be required for future experimental designs using the proposed method to detect effect size=10%, with confidence=95% and power=0.8。
我是这个问题。
【在 B******y 的大作中提到】
: 你都已经有数据了，Sample Size＝8，你还需要做什么Power Analysis？
| c****g
发帖数: 156 | 4 PASS, nQuery, SAS power and sample size...or just pen and paper. But i don't think 8 sample is large enough to represent the population. | g********r
发帖数: 8017 | 5 把这八个数的mean和sd算出来，随便找个sample size计算器，有钱买PASS，没钱网上有的是。比如http://www.stat.uiowa.edu/~rlenth/Power/
group 1 mean ＝ sample mean
group 2 mean ＝ sample mean＊1。1
equal sd, both = sample sd
alpha = 0.05
beta = 0.8
group size r＝1
然后求sample size就得了。
有包子，答的多答的好的包子大）：
% change (is this termed as effect size?) with 95% confidence (is this
termed as significance criterion?) with power of 0.8 (is this usually just
fixed at 0.8?)
来作为normal mean and standard deviation 这些标准的.
same number of participants in disease group and the control group, i.e.
sample size) that would be required for future experimental designs using
the proposed method to detect effect size=10%, with confidence=95% and power
=0.8。
sample size的?
free software也太好了。
【在 l**********o 的大作中提到】
: 我没怎么学过statitics，可能问题很傻，但现在没人可以问，大包子请教（有答就有包子，答的多答的好的包子大）：
: 我有一组8个数据，say (1.00,1.05,1.10,0.97,0.82,1.03,0.99,0.95)
: 我想做power analysis, to calculate the required sample size to detect a 10% change (is this termed as effect size?) with 95% confidence (is this termed as significance criterion?) with power of 0.8 (is this usually just fixed at 0.8?)
: 说得也许不清楚，换言之，我这8个是已测量的normal group，是一个group里的，用来作为normal mean and standard deviation 这些标准的.
: We want to determine the number of patients and controls (assuming the same number of participants in disease group and the control group, i.e. sample size) that would be required for future experimental designs using the proposed method to detect effect size=10%, with confidence=95% and power=0.8。
: 请教有什么软件可以让我输入我的8个数据和以上这些的paramters算出这个required sample size的?
: 如果谁已经有这样的code（我最熟悉matlab)能给我用就太棒了！或者别人写好的free software也太好了。
: 还有，我上面这些对问题的描述有明显很外行的错误吗？
: 多谢！
|
|
